---
title: Genomic Feature Analysis of CMR
output: 
  flexdashboard::flex_dashboard:
    orientation: rows
    social: menu
    source_code: embed
    theme: cosmo
---

```{r setup, include=FALSE}
library(flexdashboard)
library(rtracklayer)
library(plotly)
library(ggplot2)
library(seqinr)
library(GenomicFeatures)
library(knitr)
library(kableExtra)
options(scipen = 20)
knitr::opts_chunk$set(echo = TRUE,
                      warning = FALSE,
                      message = FALSE)
```


```{r echo = FALSE, warning = FALSE, message = FALSE}
library(flexdashboard)
# calculate exon, intron and gene length from txdb according to geneID
exonLength <- function(geneID, txdb){
  exonTxdb <- exonsBy(txdb, "gene")
  tmp <- exonTxdb[geneID]
  exonLen <- unlist(lapply(tmp,
                           function(x) sum(width(reduce(x)))))
  geneGR <- genes(txdb)
  geneLength <- width(geneGR)
  names(geneLength) <- geneGR$gene_id
  geneLength <- geneLength[geneID]
  resMat <- data.frame(geneID = geneID,
                       geneLength = geneLength,
                       exonLength = exonLen)
  resMat$intronLength <- resMat$geneLength - resMat$exonLength
  resMat
}

CMRGene <- read.table(file = CMRGeneDir, sep = "\t", header = F, quote = '',
                      stringsAsFactors = F)
CMRGene <- CMRGene$V1

GTF <- import(con = GTFDir)
GTF <- subset(GTF, GTF$gene_biotype == "protein_coding") # only protein coding gene are used for downstream analysis
txdb <- makeTxDbFromGFF(file = GTFDir) # from R package GenomicFeatures

seqNames <- na.omit(as.numeric(as.character(unique(seqnames(GTF)@values))))
GTFdf <- data.frame(GTF)

resFreq <- NULL
resIntronLength <- NULL
for(i in 1:length(seqNames)){
  curGTF <- subset(GTFdf, GTFdf$seqnames == i)
  exonNumber <- by(data = curGTF, INDICES = curGTF[, "gene_id"],
                   function(x) length(na.omit(unique(x[,"exon_number"]))))
  curGeneGTF <- subset(curGTF, curGTF$type == "gene")
  curGeneGTF <- curGeneGTF[order(curGeneGTF$start), ]
  rownames(curGeneGTF) <- curGeneGTF$gene_id
  orderGene <- rownames(curGeneGTF)
  exonNumber <- exonNumber[orderGene]
  curLength <- length(exonNumber)
  curStart <- seq(1, curLength - seqLength + stepSize, stepSize)
  curEnd <- c(curStart[1:(length(curStart)-1)] + seqLength - 1, curLength)
  curMat <- cbind(curStart, curEnd)
  CMRFreq <- apply(curMat, 1,
                   function(x) length(intersect(names(exonNumber)[x[1]:x[2]],
                                                CMRGene))/seqLength)
  
  tmpFeature <- exonLength(geneID = orderGene, txdb = txdb)
  curIntronLength <- apply(curMat, 1,
                         function(x) mean(tmpFeature[orderGene[x[1]:x[2]],"intronLength"]))
  resFreq <- c(resFreq, CMRFreq)
  resIntronLength <- c(resIntronLength, curIntronLength)
}
```



### Introduction
The function provides the correlation analysis of CMR genes with multiple gene features including **Mean gene length**, **Mean exon number**, **Mean GC content** and **Mean distance to adjacent gene**. To be specific, we split the maize genome in a sliding window of 100 adjacent genes with step size of 10, and calculate the frequency of CMR genes in each window (100 adjacent genes).


Row {data-height=250}
---------------------------------------------------------------

### Correlation of CMR genes with mean intron length

```{r echo = FALSE, warning=FALSE, message=FALSE}
df <- data.frame(freq = resFreq*100, 
                 intronLength = resIntronLength/1000)
p1 <- ggplot(df, aes(x = freq, y = intronLength)) + 
      geom_point(colour = "grey") + 
      geom_smooth(method = "lm", se = FALSE, colour = "red") + 
      labs(x = "Frequency of CMR genes (%)", y = "Mean gene length (1000 bp)")
ggplotly(p1)
```


### Statistics of linear regression of CMR genes with mean intron length

```{r echo = FALSE, warning = FALSE, message = FALSE}
group <- gl(n = 2, k = nrow(df), labels = c("freq", "intronLength"))
values <- c(df$freq, df$intronLength)
lmRes <- lm(values ~ group)
res.summary <- summary(lmRes)
res.table <- data.frame(Stats = c("r.squared",
                                  "adj.r.squared",
                                  "fstatistic",
                                  "sigma",
                                  "df"),
                        values = c(res.summary$r.squared,
                                   res.summary$adj.r.squared,
                                   paste(res.summary$fstatistic, collapse = " "),
                                   res.summary$sigma,
                                   paste(res.summary$df, collapse = " ")))
knitr::kable(x = res.table, align = 'c') %>%  
  kableExtra::kable_styling(position = 'center', full_width = FALSE, 
                            stripe_color = 'black', latex_options = "bordered")
```


### Anova analysis of CMR genes with mean intron length

```{r echo = FALSE, warning = FALSE, message = FALSE}
knitr::kable(x = t(anova(lmRes)), align = "c") %>%
  kableExtra::kable_styling(position = "center", full_width = FALSE,
                            stripe_color = "black", latex_options = "bordered")
```



Row {data-height=250}
---------------------------------------------------------------

```{r echo = FALSE, warning = FALSE, message = FALSE}
nonCMRGene <- setdiff(unique(GTFdf$gene_id), CMRGene)
nonCMRGene <- sample(nonCMRGene, size = ratio*length(CMRGene))

cmrFeature <- exonLength(geneID = CMRGene, txdb = txdb)
nonCMRFeature <- exonLength(geneID = nonCMRGene, txdb = txdb)
```

### Intron length difference between CMR genes and non-CMR genes
```{r echo = FALSE, warning=FALSE, message=FALSE}
cmrStat <- boxplot.stats(cmrFeature$intronLength)$stats
tmpCMR <- cmrFeature$intronLength[which(cmrFeature$intronLength <= cmrStat[5])]
noncmrStat <- boxplot.stats(nonCMRFeature$intronLength)$stats
tmpNonCMR <- nonCMRFeature$intronLength[which(nonCMRFeature$intronLength <= noncmrStat[5])]

df <- data.frame(type = factor(rep(c("CMRGene", "nonCMRGene"),
                                   c(length(tmpCMR), length(tmpNonCMR)))),
                 value = c(tmpCMR/1000, tmpNonCMR/1000))

p <- ggplot(df, aes(x=type, y=value, fill=type)) + 
  geom_violin(trim = FALSE) +
  theme(axis.title.x = element_blank(),
        #axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        legend.title = element_blank(),
        legend.position = "none")
ggplotly(p)
```

### The statistics of intron length difference between CMR genes and non-CMR genes
```{r echo = FALSE, warning=FALSE, message=FALSE}
res <- data.frame(Stats = c("Min", "First quantile", "Median", "Third quantile", "Max"),
                  CMR = cmrStat, nonCMR = noncmrStat)
knitr::kable(x = res, align = 'c') %>%  
  kableExtra::kable_styling(position = 'center', full_width = FALSE, 
                            stripe_color = 'black', latex_options = "bordered")
```



### Intron length density difference between CMR genes and non-CMR genes
```{r echo = FALSE, warning = FALSE, message = FALSE}
p <- ggplot(df, aes(x=value, fill=type)) + 
  geom_density(alpha = 0.5) +
  theme(axis.title.x = element_blank(),
        #axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        legend.title = element_blank(),
        legend.position = "none")
ggplotly(p)
```